Calibration frames - how many / what type to take

[joecoyle]

Hi,

I'm determined to start to make more of my imaging so I am trying to build up a list of all the calibration frames I should take (and probably forgot to take from day 1)... thus far I have only ever taken bias / darks with the luminance filter.

As I have a mono camera, do I need bias/darks for each colour too?

Thus far, the best sub lengths I can get, are 1 minute, 2 minute and 5 minute. So I originally went for a stack of 50x4 minutes & made a master dark, and PixInsight seems to handle scaling it up/down. No idea if it's actually any good though.

Now, I'm thinking I need to redo it - the cameras been moved around a lot in the obsy since then.

Table time...

+-----------+------+----------+----------+
|  Filter   | Type |  Length  | Quantity |
+-----------+------+----------+----------+
| Luminance | Bias | 0.0001   |      200 |
| Luminance | Dark | 1 minute |       50 |
| Luminance | Dark | 2 minute |       50 |
| Luminance | Dark | 5 minute |       50 |
| ---       |      |          |          |
| Red       | Bias | 0.0001   |      200 |
| Red       | Dark | 1 minute |       50 |
| Red       | Dark | 2 minute |       50 |
| Red       | Dark | 5 minute |       50 |
+-----------+------+----------+----------+

Am I right in this? Or totally off the mark?

Do I need to do the same for Green/Blue / all my narrowband filters?

Do I need a flat per colour / per length too? I can edit the table as we go, so I'd appreciate any help in filling it in /expanding it

Thanks

Joe

Edited April 21, 2020 by joecoyle
table

[Davey-T]

Hi Joe, good to see you're getting back into imaging.

50 darks is a bit overkill IMHO although some folks think its a good idea, I did do some experimenting in the dim distant past and the difference between 16 darks and 32 darks wasn't noticeable bearing in mind my light polluted skies, I know that mathematically you can make a case for lots of darks but didn't seem to have much effect in the real world.

I take flats  for each filter but then again there are others that don't so I guess it's a case of suck it and see, don't use PI  but Maxim will go look for matching calibration frames and apply the right ones to the right filters no idea what PI does.

Flats are the same for all length exposures.

Dave

Edited April 21, 2020 by Davey-T

[martin_h]

No need for darks or bias for coloured filters......colour doesn't count in the dark!!! One size fits all, use bias for all light frames, use darks for all light frames matching exposure times, the only time you may need coloured calibration is for flats if you have dust on your filters.

[Freddie]

You are a bit off on this.  As there is no light getting to the cam for bias and darks there is no need to do them for each filter.

For flats they are not determined by light sub duration but are determined by exposure length to get the correct histogram so you only need one for each filter. Should be done for each filter but depending on your setup you may get away with using a flat from the L filter for everything.

 

Edit: I should have typed faster!!!

Edited April 21, 2020 by Freddie

[joecoyle]

Ahhh major idiot moment. See when you say it like that it makes more sense - of course there is no light, so why would colour count - d'oh... Misconception number 2 on my list  

My brain is melted after trying to read so much around it.

Im guessing temperature does matter though - I'm cooling my cam down now and its at -13... The shed is quite warm though, so I might wait til later to get it down to -20.

I've been following Kayron's excellent tutorials at https://www.lightvortexastronomy.com/tutorials.html and whilst they are super detailed and informative, sometimes I get lost.

How many flats per filter do you take? Just 1 or several and stack them into a master flat? Kayron suggests several of each filter and making a master.

Dare I throw binning into the mix?

Cheers

Joe

Edited April 21, 2020 by joecoyle

[martin_h]

I have been taking darks all afternoon as it happens, -20 deg 2,3,5,15 mins 10 of each.

[tooth_dr]

I've a few darks and flats to take at the minute, but I've been using the good weather to image.  I normally take around 50.  At the end of the day it doesnt cost me anything to take them, and so a few extra frames wont hurt, (but there is no logic in it).

I take my darks and flats at night to eliminate light leakage as much as possible, as the back of my reflector is open.  My camera was recording a temp of 19.8oC today so it would have be very difficult to maintain a nice -20deg anyway.

I take one flat (using luminance filter) and that seems to work ok for LRGB.  You would again take as many flats frames as you want to, and make a master flat.

If you bin, then take a new set of all calibration data at the new binning level eg 2x2.

I have tried scaling my darks, it doesnt work of me, plus it is easy just to have a set for each time interval.  I keep my master darks handy and reuse for months.

Temp matters too, so calibration frames needs taken at the same temp as your lights - pick a temp and stick to it otherwise it's a PITA.

Make all your calibration frames, save them in a folder, and reuse as required, and then take more when the weather is bad or if they dont calibrate properly.  You may need to take flats regularly if you move the kit about a lot and remove the camera.  I have a permanent set up so I dont take flats often.  Which is a good thing as my flats needs to be over 3.5s in length, and that can be a right hassle.

Oh and dark flats - these are darks used to calibrate your flats.  Whatever length of time your flats turns out to be, take a corresponding set of darks (ie lens cap on) at the same length.  You can see how it you take a separate flat for each filter, you can end up with a lot of flats and darks flats.

 

HTH

Adam.

 

 

[joecoyle]

@tooth_dr brilliant thanks for the explanation -  I had totally forgotten about dark flats too 😕

Yeah I've been reusing these old darks for far too long - about time I redid them and my flat box is still in its box - a Christmas present from 2018. Been too scared to use it, as I always thought parking the mount at the end of the night meant all the dust moved around inside making flats pointless. So I never bothered as I value my sleep too much to be taking them at 3am when the mount is still tracking the target

Now I have a permanent setup now with my mount on a concrete pier and I have no plans to rotate my camera anytime soon, so I should be able to take flats the following day do you think? Just point the scope up and rest the flat panel on it? I'm told as long as the camera orientation does not change, I should be good to go...

Thanks

Joe

[tooth_dr]

11 minutes ago, joecoyle said:

Now I have a permanent setup now with my mount on a concrete pier and I have no plans to rotate my camera anytime soon, so I should be able to take flats the following day do you think? Just point the scope up and rest the flat panel on it? I'm told as long as the camera orientation does not change, I should be good to go...

 

Yes that's 100%.  Just watch that light cant get in anywhere as this can affect the flats/darks.

[ollypenrice]

Joe, your camera is CCD rather than CMOS so you do not need to shoot darks for flats. Instead you can use a master bias as a dark for your flats and save yourself a lot of bother. At the short exposures times of flats there will be no significant difference between a dark of that short exposure time and a bias, so just use a master bias as a flat dark. It may help to make a copy of your master bias and re-name it 'Flat dark' or something because some stacking software doesn't like seeing the same file name in two places.

Olly

Edit: CMOS cameras do need flat darks taken at the same settings as the flats.

 

Edited April 22, 2020 by ollypenrice

[iapa]

On 21/04/2020 at 19:06, martin_h said:

I have been taking darks all afternoon as it happens, -20 deg 2,3,5,15 mins 10 of each.

I only cool to -5C based on the camera's thermal noise graph below

If my thermal noise is much less than my light pollution noise, and it's actually much less than 1% of LP, then, really, I don't think I need to waste the energy to cool the camera at all as it is such a tiny fraction of the total noise.

I do cool to -5C, just to give me a constant temp.

Although, in the winter, I may actually need to heat the camera  when it gets to <-15C or lower

That's one for me to work on over the next few months before I get any sunlight free skies.

 

[rickwayne]

While darks can be scaled to different exposure times and temperatures, that process is not perfect, and you may as well just build yourself a set of them. With a cooled camera, you can do this at any time; folk who lack those often wind up spending imaging time on in-situ darks. So, you know, yay you! Personally I shoot them in sets of 100 because hey, why not? Once I've built the sequences, which takes a minute or two, the camera and computer will happily shoot for days to complete the library. Once I've averaged master darks out of the individual frames, I can throw the latter away. The complexity of managing all these calibration frames is a strong argument for settling on a set of parameters (gain, bias, temperature, exposure time) that work for you and then just reusing that as much as possible.

Again, you may find that your flats work between filters. But dust happens -- I just reshoot every time. I have a sequence in Ekos for that too, so I just have to trigger it while I'm tearing down and packing up (and the 11x14 tracing pad makes a nice little lamp for that process, as I've said).

Either flat dark frames or bias frames are required for the flats math to work out correctly. Anecdotes about super-short exposures leading to banding and other problems on certain CMOS cameras lead many to avoid them; imagers I respect say they see no evidence of such problems. I have some small measure of self-respect, I guess, and I sure don't. The advantage to bias frames over flat darks is that the latter must be the same exposure time as your flats; if you are shooting per-filter flats correctly, you're 99% certain to be shooting different exposures for OIII and luminance! One set of bias frames works for all exposures. Simple.

Finally, as with all noise-reduction techniques, cooling the camera is a diminishing-returns game. You do you, but cooling to -20 gives a pretty marginal improvement at best over higher temperatures. Here I am flogging Robin Glover's talk yet again, but IMO everybody should give it a squint, and figure out where the bang/buck peaks for them. CCD fellers likely colder than us CMOS types, stipulated. 

 

[iapa]

@rickwayne, that presentation is where i got the idea to not try to "super" cool the camera.

Main reason for -5C, is that from about October I can generally expect that over night until late March.

Or into May this year 

As I have time, I am taking a set of darks/dark flats, bias for 30,60,90,120,300 and 600 secs - just to give me a library that I can use to do comparative testing, nothing much better to do just now.

Apart from using utoob videos for trying to work out how to use PixInsight and CCD Inspector that is.

Edited May 22, 2020 by iapa